CN113836812B - Shield construction pose adjusting method for identifying hard rock thickness by using intelligent algorithm - Google Patents

Shield construction pose adjusting method for identifying hard rock thickness by using intelligent algorithm Download PDF

Info

Publication number
CN113836812B
CN113836812B CN202111146030.XA CN202111146030A CN113836812B CN 113836812 B CN113836812 B CN 113836812B CN 202111146030 A CN202111146030 A CN 202111146030A CN 113836812 B CN113836812 B CN 113836812B
Authority
CN
China
Prior art keywords
hard rock
decision tree
thickness
data
sample set
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202111146030.XA
Other languages
Chinese (zh)
Other versions
CN113836812A (en
Inventor
姜泓任
姜谙男
金圣珂
虢新平
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Dalian Maritime University
Original Assignee
Dalian Maritime University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Dalian Maritime University filed Critical Dalian Maritime University
Priority to CN202111146030.XA priority Critical patent/CN113836812B/en
Publication of CN113836812A publication Critical patent/CN113836812A/en
Application granted granted Critical
Publication of CN113836812B publication Critical patent/CN113836812B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • E—FIXED CONSTRUCTIONS
    • E21—EARTH OR ROCK DRILLING; MINING
    • E21D—SHAFTS; TUNNELS; GALLERIES; LARGE UNDERGROUND CHAMBERS
    • E21D9/00—Tunnels or galleries, with or without linings; Methods or apparatus for making thereof; Layout of tunnels or galleries
    • E21D9/06—Making by using a driving shield, i.e. advanced by pushing means bearing against the already placed lining
    • E—FIXED CONSTRUCTIONS
    • E21—EARTH OR ROCK DRILLING; MINING
    • E21D—SHAFTS; TUNNELS; GALLERIES; LARGE UNDERGROUND CHAMBERS
    • E21D9/00—Tunnels or galleries, with or without linings; Methods or apparatus for making thereof; Layout of tunnels or galleries
    • E21D9/06—Making by using a driving shield, i.e. advanced by pushing means bearing against the already placed lining
    • E21D9/093—Control of the driving shield, e.g. of the hydraulic advancing cylinders
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N10/00—Quantum computing, i.e. information processing based on quantum-mechanical phenomena
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/10—Numerical modelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Mining & Mineral Resources (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computing Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • Medical Informatics (AREA)
  • Geometry (AREA)
  • Geology (AREA)
  • Computer Hardware Design (AREA)
  • Geochemistry & Mineralogy (AREA)
  • General Life Sciences & Earth Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Environmental & Geological Engineering (AREA)
  • Condensed Matter Physics & Semiconductors (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Health & Medical Sciences (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a shield construction pose adjusting method for identifying hard rock thickness by using an intelligent algorithm, which comprises the following steps of: s1: establishing an orthogonal design scheme and a uniform design scheme; s2: establishing a numerical model of stratum and shield construction, and calculating the original muck density; s3: constructing a data sample set I and a data sample set II; s4: preprocessing data in the data sample set I and the data sample set II; s5: training a multi-decision tree model; s6: optimizing the number N of decision trees and the number S of features in the feature subsets: s7: forming a multi-decision tree prediction model; s8: identifying the intrusion thickness of the hard rock through a multi-decision tree model; and outputting shield construction pose adjustment parameters. The method utilizes numerical simulation to obtain the corresponding relation between the downward movement and downward inclination of the shield construction axis and the intrusion thickness of the hard rock, and utilizes the ratio of the elastic modulus of the hard rock and the soft rock to correct parameters so as to obtain reasonable shield construction pose control parameters.

Description

Shield construction pose adjusting method for identifying hard rock thickness by using intelligent algorithm
Technical Field
The invention relates to the field of digital processing analysis, in particular to a shield construction pose adjusting method for identifying hard rock thickness by using an intelligent algorithm.
Background
In recent years, with the rapid increase of urban economy in China, the construction of subways begins to enter a large development period. Most of large cities are planning and constructing own rail transit, and quite a lot of cities adopt a shield method to carry out subway construction. The shield construction is an automatic control system, but has the adaptability problem to geological environment, that is, different shield construction parameters such as thrust, torque, slag output and water injection are required to be adopted for different geological environments. Particularly, when a composite stratum exists and the soft and hard strata are mixed and complicated, the shield body rises and rises due to the invasion of the hard rock at the lower part when the shield is driven, so that great deviation is brought to construction.
The hard rock invading into the soft rock stratum can firstly cause the 'head-up' phenomenon of the shield machine, namely the lifting phenomenon of the shield machine by the hard rock has great relation with the different invading thicknesses of the hard rock. In actual construction, the hard rock invasion depth of a composite stratum cannot be directly observed, only the experience of shield machine operators can be used for roughly judging, and great human influence factors exist, so that the normal construction of a shield is influenced.
Disclosure of Invention
The invention provides a shield construction pose adjusting method for identifying hard rock thickness by using an intelligent algorithm, which aims to solve the technical problem that the normal construction of a shield is influenced by a great human factor because the invasion depth of hard rock of a composite stratum cannot be directly observed in actual construction and can only be roughly judged by the experience of shield machine operators.
A shield construction pose adjusting method for identifying hard rock thickness by using an intelligent algorithm comprises the following steps:
s1: establishing an orthogonal design scheme and a uniform design scheme; the orthogonal design scheme and the uniform design scheme are parameter combinations with the invasion thickness of the hard rock of the stratum, the elastic modulus of the soft rock stratum, the elastic modulus of the hard rock stratum and the Poisson ratio as parameters;
s2: establishing a numerical model of stratum and shield construction to obtain the results of crown arch settlement displacement, bottom plate rising displacement and lateral convergence displacement of a typical section corresponding to the parameters of the orthogonal design scheme and the parameters of the uniform design scheme, and calculating the original muck density corresponding to the orthogonal design scheme and the uniform design scheme;
s3: constructing a data sample set I of the orthogonal design scheme and a data sample set II of the uniform design scheme, wherein the data sample set I and the data sample set II respectively comprise the top arch settlement displacement, the bottom plate rising displacement, the lateral convergence displacement, the original muck density, the stratum hard rock invasion thickness, the soft rock elastic modulus and the hard rock elastic modulus; the top arch settlement displacement, the bottom plate rising displacement, the lateral convergence displacement and the original muck density are used as input parameters of the data sample set I and the data sample set II, and the stratum hard rock invasion thickness, the soft rock elastic modulus and the hard rock elastic modulus are used as output parameters of the data sample set I and the data sample set II;
s4: preprocessing the data in the data sample set I and the data sample set II to normalize the data to obtain a normalized data sample set I and a normalized data sample set II;
s5: training a multi-decision tree model through the data sample set I after normalization processing to obtain a mapping relation between input and output of the multi-decision tree model;
s6: optimizing the number N of decision trees and the number S of features in the feature subset in the multi-decision tree model by a quantum fruit fly algorithm: obtaining the number N of decision trees in the optimal multi-decision tree model and the number S of features in the feature subset;
s61: initializing the fruit fly algorithm parameters;
s62: adjusting the random direction and the step length of the fruit fly algorithm to update the optimizing route of the fruit fly algorithm;
s63: aiming at the optimization variable corresponding to each generation of fruit flies, training and predicting the multi-decision tree model through the data sample set I after normalization processing; calculating the adaptive value of each generation of fruit flies according to the adaptive value function; predicting the multi-decision tree model through a data sample set II after normalization processing to obtain an adaptive value function;
s64: sorting the adaptation values of the fruit flies in the current generation, and iterating the fruit flies in the size of the elite fruit flies to obtain fruit fly individuals with optimal fitness;
s65: storing the adaptive value and the position coordinate of the current optimal drosophila individual;
s66: if the current fruit fly algorithm reaches the maximum iteration times, outputting the optimal value of the current fruit fly, otherwise, repeating the steps from S62 to S64;
s7: training the obtained optimal parameters of the drosophila algorithm through a data sample set I after normalization processing to form a multi-decision tree prediction model;
s8: identifying the intrusion thickness of the hard rock through a multi-decision tree model; and then, carrying out shield pose adjustment according to the intrusion thickness of the hard rock, and outputting shield construction pose adjustment parameters, wherein the shield construction pose adjustment parameters comprise downward translation quantity delta and downward inclination angle beta.
Further, in S4, the method for normalizing the data includes:
Figure BDA0003285578770000031
wherein: y is max The maximum value of the output column of the data set; y is min Is the minimum of the output column of the data set; x is the number of max Is the maximum value of the input column of the data set; x is the number of min Is the minimum value of the input column of the data set; xb is the normalized value of the sample input column; yb is the normalized value of the sample output column; x is the original data of the sample input column; y is the raw data for the sample output column.
Further, in S5, the multi-decision tree model is trained as follows:
Y＝RM(X) (2)
Y＝[yb 1 ,yb 2 ,…,yb i ,…,yb l ],X＝[xb 1 ,xb 2 ,…,xb j ,…,xb m ].
in the formula: y is the normalized output data, l is the dimension (column) of the output data, X is the normalized input data, and m is the dimension (column) of the input data; RM stands for multi-decision tree model.
Further, the method for initializing the drosophila algorithm parameters in S61 is as follows: setting the fruit fly population scale as P; the size of the Elaphanita elite fruit fly is Z; the maximum number of iterations of the Drosophila algorithm is n max Taking the number N of the decision trees and the number S of the features in the feature subsets as optimization variables of the drosophila algorithm, namely X i ＝[N i ,S i ]Wherein i is the number corresponding to the fruit fly in the population; the initial fruit fly positions were randomly generated.
Further, the random direction and step size of the drosophila algorithm are adjusted in S62 according to an adaptive step size strategy, and the optimization route for updating the drosophila algorithm is as follows:
Figure BDA0003285578770000032
Figure BDA0003285578770000041
{X i ＝X axis +α*RandomValue
X axis the initial value of the optimized variable is obtained;
Figure BDA0003285578770000042
the average value of the corresponding adaptive values of P fruit flies is obtained; n represents the nth fruit fly; func (X) i ) Is X i An adaptive value function; the R band represents the rate of change of population average odor concentration; alpha is alphaRepresents an update weight; RandomValue represents a search direction and distance.
Further, in S63, the mapping relationship in the multi-decision tree model is predicted through the data sample set II after the normalization processing, and the adaptive value function is obtained as follows:
Figure BDA0003285578770000043
in the formula: wherein yb' h Predicting the test sample values, yb, for a multi-decision tree h Actual test sample data values; f MSE Root mean square error for a plurality of test sample points; m is the number of the sample data to be tested, and h is the h-th sample.
Further, in S64, the method for iterating the drosophila melanogaster discharged into the elite drosophila size is to introduce quantum operation, and the iterative process of calculating according to the following formula is performed according to the quantum operation of the drosophila location as follows:
Figure BDA0003285578770000044
Figure BDA0003285578770000045
in the formula: n represents the nth fruit fly:
Figure BDA0003285578770000046
and u are both random numbers between (0, 1); λ is an innovation parameter;
Figure BDA0003285578770000047
the local optimal value of the nth generation of the ith fruit fly;
Figure BDA0003285578770000048
global optimal solution for nth generation of all fruit flies;
Figure BDA0003285578770000049
represents the average of the local optimal solutions of the population of the nth generation of Elaphanita eligua.
Further, the method for flying all drosophila individuals in S65 to the optimal drosophila individuals is as follows:
Figure BDA00032855787700000410
Figure BDA00032855787700000411
X axis ＝X(BestIndex)；
in the formula, ibestIndex is the serial number of the current optimal solution of the ith fruit fly; BestIndex is the individual number of the currently best Drosophila within the population.
Further, the training process of forming the multi-decision tree prediction model in S7 is as follows:
s71: randomly and repeatedly extracting u samples from an original learning set, and carrying out N times of sampling to form N learning sample subsets;
s72: for the N learning sample subsets, developing N decision trees;
s73: for each decision tree, assuming that S characteristic attributes are in total, selecting an optimal attribute as a root node of the decision tree, and then splitting in sequence;
s74: splitting each decision tree according to the optimal attribute to form a multi-decision tree;
s75: and averaging according to the regression result of each decision tree to obtain the mapping relation between the input and the output of the multi-decision tree model.
Further, in S8, the shield pose adjustment process according to the hard rock intrusion thickness is as follows:
s81: inputting the monitoring displacement and the muck density into a trained model of the multi-decision tree;
s82: the trained model of the multi-decision tree outputs hard rock invasion thickness, soft rock elastic modulus and hard rock elastic modulus;
s83: the hard rock invasion thickness is corrected as follows:
θ＝E 1 /E 2 (6)
H′＝H×δ (7)
in the formula, E 1 Is the modulus of elasticity of hard rock, E 2 Is the elastic modulus of the soft rock, theta is the elastic modulus ratio of the hard rock and the soft rock, and delta is the correction coefficient of the invasion thickness of the hard rock; h' is the corrected hard rock invasion thickness, and H is the hard rock invasion thickness predicted by the multi-decision tree.
The invention discloses a shield construction pose adjusting method for identifying hard rock thickness by using an intelligent algorithm.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a schematic diagram of the shield tunneling machine head-up phenomenon of invasion of stratum hard rock into a composite stratum according to the invention;
FIG. 2 is a flow chart of a shield construction pose adjustment method for identifying hard rock thickness by using an intelligent algorithm according to the invention;
FIG. 3 is a schematic diagram of a multi-decision tree training process according to the present invention;
FIG. 4 is a schematic diagram illustrating a process for invoking a multi-decision tree model for prediction according to the present invention;
FIG. 5a is a schematic view of the downward translation control operation of the shield tunneling machine according to the present invention;
FIG. 5b is a schematic view of the downward inclination control operation of the shield tunneling machine according to the present invention;
FIG. 6 is a flow chart of shield pose adjustment using formation hard rock intrusion thickness according to the present invention;
FIG. 7a is a three-dimensional numerical model of the invention when the formation hard rock is embedded at 0 m;
FIG. 7b is a three-dimensional numerical model of the present invention when the formation hard rock is embedded for 2 m;
FIG. 7c is a three-dimensional numerical model of the present invention with hard formation rock embedded for 4 m;
FIG. 7d is a three-dimensional numerical model of the invention with formation hard rock embedded for 6 m;
FIG. 8 is a diagram of the arrangement of the measuring points of the shield tunnel according to the present invention;
FIG. 9a is a vertical displacement diagram of the invention with hard rock embedded at 0 m;
FIG. 9b is a vertical displacement diagram of the invention with formation hard rock embedded for 2 m;
FIG. 9c is a vertical displacement diagram of the formation hard rock embedded at 4m according to the invention;
FIG. 9d is a vertical displacement plot of a 6m hard rock formation according to the present invention;
FIG. 10a is a diagram of the maximum principal stress of the shield tunneling machine at different inclination angles when the stratum hard rock is embedded into the stratum hard rock of the invention at 0 m;
FIG. 10b is a diagram of the maximum principal stress of the shield tunneling machine at different inclination angles when the stratum hard rock is embedded for 2m according to the invention;
FIG. 10c is a graph of the maximum principal stress at different dip angles of the shield tunneling machine when the formation hard rock is embedded for 4m according to the present invention;
FIG. 10d is a diagram of the maximum principal stresses at different dip angles of the shield tunneling machine when the formation hard rock is embedded for 6m according to the present invention;
FIG. 11a is a corresponding relationship curve between the intrusion thickness of the shield tunneling machine along with the formation hard rock and the shield declination angle;
FIG. 11b is a corresponding relation curve between the intrusion thickness of the shield tunneling machine along with the formation hard rock and the downward movement amount of the shield tunneling machine.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention provides a shield construction pose adjusting method for identifying hard rock thickness by using an intelligent algorithm, which is used for preventing the problem of out-of-control shield pose caused by the variation of the upper soft reflection and the lower reflection of a stratum, such as a shield machine head-up phenomenon schematic diagram of a stratum hard rock invading a composite stratum shown in an attached figure 1; the adjusting method is shown in the attached figure 2 and comprises the following steps:
s1: establishing an orthogonal design scheme and a uniform design scheme; the orthogonal design scheme and the uniform design scheme are parameter combinations with the invasion thickness of the hard rock of the stratum, the elastic modulus of the soft rock stratum, the elastic modulus of the hard rock stratum and the Poisson ratio as parameters;
s2: and establishing a numerical model of stratum and shield construction to obtain the results of crown arch settlement displacement, bottom plate rising displacement and lateral convergence displacement of the typical section corresponding to the parameters of the orthogonal design scheme and the parameters of the uniform design scheme, and calculating the original muck density corresponding to the orthogonal design scheme and the uniform design scheme. The numerical models of the stratum and the shield construction are input by combining parameters of the orthogonal design scheme and the uniform design scheme;
s3: constructing a data sample set I of the orthogonal design scheme and a data sample set II of the uniform design scheme, wherein the data sample set I and the data sample set II respectively comprise the top arch settlement displacement, the bottom plate rising displacement, the lateral convergence displacement, the original muck density, the stratum hard rock invasion thickness, the soft rock elastic modulus and the hard rock elastic modulus; the top arch settlement displacement, the bottom plate rising displacement, the lateral convergence displacement and the original muck density are used as input parameters of the data sample set I and the data sample set II, and the stratum hard rock invasion thickness, the soft rock elastic modulus and the hard rock elastic modulus are used as output parameters of the data sample set I and the data sample set II;
s4: preprocessing the data in the data sample set I and the data sample set II to normalize the data to obtain a normalized data sample set I and a normalized data sample set II; in this embodiment, the data is normalized to (0,1), and the specific method is as follows:
Figure BDA0003285578770000071
wherein: y is max The maximum value of the output column of the data set; y is min Is the minimum of the output column of the data set; x is the number of max Is the maximum value of the input column of the data set; x is the number of min Is the minimum of the input column of the data set; xb is the normalized value of the sample input column; yb is the normalized value of the sample output column; x is the original data of the sample input column; y is the raw data for the sample output column.
S5: training a multi-decision tree model through the data sample set I after normalization processing to obtain a mapping relation between input and output of the multi-decision tree model; the method comprises the following specific steps:
Y＝RM(X) (2)
Y＝[yb 1 ,yb 2 ,…,yb i ,…,yb l ],X＝[xb 1 ,xb 2 ,…,xb j ,…,xb m ].
in the formula: y is the normalized output data, l is the dimension (column) of the output data, X is the normalized input data, and m is the dimension (column) of the input data; RM stands for multi-decision tree model.
Specifically, the RM model is composed of a plurality of decision trees, where two parameters, the number N of the decision trees and the number S of features in the feature subset, have a relatively large influence on the accuracy of the mapping prediction, and the 2 parameters are optimized by using the quantum drosophila algorithm.
S6: optimizing the number N of decision trees in the multi-decision tree model and the number S of features in the feature subsets through a quantum fruit fly algorithm to obtain the number N of decision trees in the optimal multi-decision tree model and the number S of features in the feature subsets;
s61: the method for initializing the fruit fly algorithm parameters comprises the following steps:
setting the fruit fly population scale as P, namely setting the quantity of fruit flies as P; the size of the Elaphe odorata fruit flies is Z, namely the number of the Elaphe odorata fruit flies is Z, and the maximum iteration number of the fruit fly algorithm is n max Taking the number N of the decision trees and the number S of the features in the feature subsets as optimization variables of the drosophila algorithm, namely X i ＝[N i ,S i ]Wherein i is the number corresponding to the fruit fly in the population; the initial fruit fly positions were randomly generated.
S62: adjusting the random direction and the step length of the fruit fly algorithm to update the optimizing route of the fruit fly algorithm; in this embodiment, the random direction and step size of the drosophila algorithm are adjusted based on an adaptive step size strategy, and the optimization route for updating the drosophila algorithm is as follows:
Figure BDA0003285578770000081
X axis the initial value of the optimized variable is obtained;
Figure BDA0003285578770000082
the average value of the corresponding adaptive values of P fruit flies is obtained; n represents the nth fruit fly; func (X) i ) Is X i An adaptive value function; the R band represents the rate of change of population average odor concentration; α represents an update weight; RandomValue represents a search direction and distance.
S63: aiming at the optimization variable corresponding to each generation of fruit flies, training and predicting the multi-decision tree model, namely a formula (2), through the data sample set I after normalization processing; calculating the adaptive value of each generation of fruit flies according to the adaptive value function; predicting the multi-decision tree model through a data sample set II after normalization processing;
predicting the mapping relation in the formula (2) through the data sample set II after the normalization processing, and obtaining the adaptive value function as follows:
Figure BDA0003285578770000091
in the formula: wherein yb' h Predicting the test sample values, yb, for a multi-decision tree h Actual test sample data values; f MSE Root mean square error for a plurality of test sample points; m is the number of the test sample data, and h is the h-th sample;
s64: sorting the adaptation values of the fruit flies in the current generation, and iterating the fruit flies in the size of the elite fruit flies to obtain fruit fly individuals with optimal fitness;
in this embodiment, the method for iterating the drosophila that is discharged into the size of elaiopsis elimiantes is to introduce quantum operation, and the iteration process of calculating the quantum operation according to the position of the drosophila according to the following formula is as follows:
Figure BDA0003285578770000092
Figure BDA0003285578770000093
in the formula: n represents the fruit fly of the nth generation,
Figure BDA0003285578770000094
and u are random numbers between (0,1), and the probability of taking the positive and negative is 0.5; λ is an innovative parameter and is also the only control parameter;
Figure BDA0003285578770000095
the local optimal value of the nth generation of the ith fruit fly;
Figure BDA0003285578770000096
global optimal solution for nth generation of all fruit flies;
Figure BDA0003285578770000097
represents the average value of the local optimal solution of the nth generation of Elaphanita elite fly population;
and (3) performing secondary optimization on the drosophila elite after quantum operation is performed on the drosophila elite, and finding out the drosophila individual with the optimal fitness according to the formula (4).
S65: the adaptive value and the position coordinates of the current optimal fruit fly individual are stored, and all the fruit fly individuals fly to the optimal fruit fly individual through visual positioning; the method specifically comprises the following steps:
Figure BDA0003285578770000098
Figure BDA0003285578770000101
X axis ＝X(BestIndex)；
in the formula, ibestIndex is the serial number of the current optimal solution of the ith fruit fly; BestIndex is the individual number of the current optimal fruit fly in the population;
s66: if the current fruit fly algorithm reaches the maximum iteration number, outputting the optimal value Xb of the current fruit fly as (Nb, Sb), otherwise, repeating the steps from S7 to S9;
s7: training the obtained optimal parameters of the drosophila algorithm through a data sample set I after normalization processing to form a multi-decision tree prediction model;
in this embodiment, the process of training the multi-decision tree prediction model is as follows:
substituting the optimal parameters N and S of the drosophila algorithm into the multi-decision tree model, wherein the training process of the multi-decision tree model is as follows, as shown in the attached figure 3:
s71: randomly and repeatedly extracting u samples from an original learning set, and carrying out N times of sampling to form N learning sample subsets; the samples that are being drawn are called in-bag data, and the data that are not being drawn are called out-bag data;
s72: for the N subsets of learning samples, developing N decision trees; because the samples are selected randomly, all the decision trees are independent from one another;
s73: for each decision tree, assuming that S characteristic attributes are in total, selecting an optimal attribute as a root node of the decision tree by using an information gain rate, and then splitting in sequence;
s74: each decision tree is split according to the optimal attribute, pruning is not needed in the process, and a multi-decision tree is formed;
s75: and averaging according to the regression result of each decision tree to obtain the mapping relation between the input and the output of the multi-decision tree model.
S8: inputting field monitoring displacement and muck density, and identifying the intrusion thickness of the hard rock through a multi-decision tree model; and then, carrying out shield pose adjustment according to the intrusion thickness of the hard rock, and outputting shield construction pose adjustment parameters, wherein the shield construction pose adjustment parameters comprise downward translation quantity delta and downward inclination angle beta.
Specifically, the model for multi-decision tree training of shield hard rock invasion thickness takes on-site monitored displacement and muck density as input, takes hard rock invasion thickness, soft rock elastic modulus and hard rock elastic modulus as output, and the process of calling the multi-decision tree model for prediction is shown in fig. 4.
Preferably, the operation mode of shield pose adjustment according to the hard rock invasion thickness is as follows:
two operation modes for shield pose adjustment after the hard rock at the bottom is invaded are mainly adopted, namely downward translation operation and downward rotation operation, as shown in fig. 5, and the downward translation amount delta and the downward inclination angle beta need to be determined according to the ratio of the invasion thickness of the hard rock and the elasticity modulus of the hard and soft rock.
In this embodiment, a shield pose adjustment process according to the intrusion thickness of hard rock is shown in fig. 6, which is specifically as follows:
s81: inputting the monitoring displacement and the muck density into a trained model of the multi-decision tree;
s82: the trained model of the multi-decision tree outputs hard rock invasion thickness, soft rock elastic modulus and hard rock elastic modulus;
s83: the hard rock invasion thickness is corrected as follows:
θ＝E 1 /E 2 (6)
H′＝H×δ (7)
in the formula, E 1 Is the modulus of elasticity of hard rock, E 2 Is the elastic modulus of the soft rock, theta is the elastic modulus ratio of the hard rock and the soft rock, and delta is the correction coefficient of the invasion thickness of the hard rock; h' is the corrected hard rock invasion thickness, and H is the hard rock invasion thickness predicted by the multi-decision tree.
Preferably, an orthogonal design scheme and a uniform design scheme are established, wherein the invasion thickness of the hard rock of the stratum, the elastic modulus of the soft rock, the elastic modulus of the hard rock and the Poisson ratio are parameter combinations, and numerical calculation is carried out to obtain a sample set shown in tables 1 and 2; according to the table 1 and the table 2, three-dimensional numerical simulation of hard rock intrusion shield construction is carried out, as shown in the attached figures 8-11; a finite element model is established by using ABAQUS numerical simulation software, the size of a shield tunnel model in the direction X, Y, Z is set to be 60m multiplied by 50m according to the requirements of field engineering, solid building blocks are adopted for soil body simulation, a mature Moore-Coulomb constitutive model is selected, a solid elastic shell model is selected for lining and duct pieces, the thickness of the lining duct pieces is 0.3m, and the tunnel excavation radius is 2.4 m. The displacement boundary conditions are as follows: the model applies x-direction constraint on the left side and the right side of the x-axis direction; the model applies y-direction constraint on the front side and the rear side of the y-axis direction; applying vertical constraint in the z direction to the bottom surface of the model; the model is subjected to grid division and is divided into 55680 units and 60146 nodes, and the three-dimensional model is shown in FIG. 7.
According to the calculation result of the numerical simulation, the maximum sedimentation value is positioned at the monitoring position of the centers of the two tunnels, namely right above the openings of the two tunnels, and the sedimentation amount is gradually reduced towards the two sides of the tunnels, so that the approximate W-shaped ground surface sedimentation is formed. And obtaining tables 1 and 2 according to the numerical calculation result and the residue soil density calculation result.
TABLE 1 orthogonal scheme and calculated Displacement
Figure BDA0003285578770000121
Compared with the orthogonal test, the uniform design has less test times, pays more attention to the dispersity of the test data and reduces a large amount of workload. Therefore, it is suitable as a sample to verify and fit the results of the orthogonal experiment. Scheme design was performed according to the uniform design software, and the parameters were selected as in table 2 below. And establishing a five-factor five-level uniform table for verifying the accuracy of the fitting equation of the orthogonal test. The homogeneity test was brought into the numerical model for calculation, data was extracted and the results filled in table 2.
TABLE 2 homogeneous test
Figure BDA0003285578770000122
Figure BDA0003285578770000131
And (3) constructing a decision tree model by using the data in the table 1 as a training sample and the data in the table 2 as a test sample according to the method, and mapping the nonlinear relation between the rock stratum parameters and the displacement and stress.
The inverse analysis algorithm adopts a quantum fruit fly-multi-decision tree algorithm, relevant parameters are set in the algorithm, and data are brought into software for learning and calculation to obtain relevant results.
TABLE 3 Back-analysis results of the homogeneous test protocol for the surrounding rock parameters
Figure BDA0003285578770000132
As can be seen from table 3, the error rate of the results of the back analysis was 8.0% at the maximum and 0.5% at the minimum. The errors of the rest data are positioned between the two data, so that the data analysis method can meet the precision requirement of actual construction and can be applied to the calculation of parameter inverse analysis.
Taking data obtained by actual engineering detection as a control value, and carrying the control value into software to calculate, wherein the vault settlement value AZ is 9.73mm, the arch bottom uplift value BZ is 7.47mm, the horizontal convergence value DZ is 8.58mm, and the density value of the residue soil is 2082kg/m 3 The results of the optimized inverse analysis are shown in Table 4.
TABLE 4 Back analysis results of the surrounding rock parameters
Figure BDA0003285578770000133
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.

Claims (10)

1. A shield construction pose adjusting method for identifying hard rock thickness by using an intelligent algorithm is characterized by comprising the following steps:
s1: establishing an orthogonal design scheme and a uniform design scheme; the orthogonal design scheme and the uniform design scheme are parameter combinations with the invasion thickness of the hard rock of the stratum, the elastic modulus of the soft rock stratum, the elastic modulus of the hard rock stratum and the Poisson ratio as parameters;
s2: establishing a numerical model of stratum and shield construction to obtain the results of crown settlement displacement, bottom plate rising displacement and lateral convergence displacement of a typical section corresponding to the parameters of the orthogonal design scheme and the parameters of the uniform design scheme, and calculating the original muck density corresponding to the orthogonal design scheme and the uniform design scheme;
s3: constructing a data sample set I of the orthogonal design scheme and a data sample set II of the uniform design scheme, wherein the data sample set I and the data sample set II respectively comprise the top arch settlement displacement, the bottom plate rising displacement, the lateral convergence displacement, the original muck density, the stratum hard rock invasion thickness, the soft rock elastic modulus and the hard rock elastic modulus; the top arch settlement displacement, the bottom plate rising displacement, the lateral convergence displacement and the original muck density are used as input parameters of the data sample set I and the data sample set II, and the stratum hard rock invasion thickness, the soft rock elastic modulus and the hard rock elastic modulus are used as output parameters of the data sample set I and the data sample set II;
s4: preprocessing the data in the data sample set I and the data sample set II to normalize the data to obtain a normalized data sample set I and a normalized data sample set II;
s5: training a multi-decision tree model through the data sample set I after normalization processing to obtain a mapping relation between input and output of the multi-decision tree model;
s6: optimizing the number N of decision trees and the number S of features in the feature subset in the multi-decision tree model by a quantum fruit fly algorithm: obtaining the number N of decision trees in the optimal multi-decision tree model and the number S of features in the feature subset;
s61: initializing the fruit fly algorithm parameters;
s62: adjusting the random direction and the step length of the fruit fly algorithm to update the optimizing route of the fruit fly algorithm;
s63: aiming at the optimization variable corresponding to each generation of fruit flies, training and predicting the multi-decision tree model through the data sample set I after normalization processing; calculating the adaptive value of each generation of fruit flies according to the adaptive value function; predicting the multi-decision tree model through a data sample set II after normalization processing to obtain an adaptive value function;
s64: sorting the adaptation values of the fruit flies in the current generation, and iterating the fruit flies in the size of the elite fruit flies to obtain fruit fly individuals with optimal fitness;
s65: storing the adaptive value and the position coordinate of the current optimal drosophila individual;
s66: if the current fruit fly algorithm reaches the maximum iteration times, outputting the optimal value of the current fruit fly, otherwise, repeating the steps from S62 to S64;
s7: training the obtained optimal parameters of the drosophila algorithm through a data sample set I after normalization processing to form a multi-decision tree prediction model;
s8: identifying the intrusion thickness of the hard rock through a multi-decision tree model; and then, carrying out shield pose adjustment according to the intrusion thickness of the hard rock, and outputting shield construction pose adjustment parameters, wherein the shield construction pose adjustment parameters comprise downward translation quantity delta and downward inclination angle beta.
2. The shield construction pose adjustment method for identifying the hard rock thickness by using the intelligent algorithm as claimed in claim 1, wherein the method for normalizing the data in S4 is as follows:
Figure FDA0003285578760000021
wherein: y is max The maximum value of the output column of the data set; y is min Is the minimum of the output column of the data set; x is the number of max Is the maximum value of the input column of the data set; x is the number of min Is the minimum of the input column of the data set; xb is the normalized value of the sample input column; yb is the normalized value of the sample output column; x is the original data of the sample input column; y is the raw data for the sample output column.
3. The shield construction pose adjustment method for identifying the hard rock thickness by using the intelligent algorithm as claimed in claim 2, wherein the training of the multi-decision tree model in S5 is as follows:
Y＝RM(X) (2)
Y＝[yb 1 ,yb 2 ,…,yb i ,…,yb l ],X＝[xb 1 ,xb 2 ,…,xb j ,…,xb m ].
in the formula: y is the normalized output data, l is the dimension (column) of the output data, X is the normalized input data, and m is the dimension (column) of the input data; RM stands for multi-decision tree model.
4. The shield construction pose adjustment method for identifying the thickness of hard rock by using the intelligent algorithm as claimed in claim 3, wherein the method for initializing the drosophila algorithm parameters in S61 is as follows:
setting the fruit fly population scale as P; the size of the Elaphanita elite fruit fly is Z; the maximum number of iterations of the Drosophila algorithm is n max Taking the number N of the decision trees and the number S of the features in the feature subsets as optimization variables of the drosophila algorithm, namely X i ＝[N i ,S i ]Wherein i is the number corresponding to the fruit fly in the population; the initial fruit fly positions were randomly generated.
5. The shield construction pose adjustment method for identifying the thickness of the hard rock by using the intelligent algorithm as claimed in claim 4, wherein the random direction and the step length of the drosophila algorithm are adjusted in S62 according to an adaptive step length strategy, and the optimization route for updating the drosophila algorithm is as follows:
Figure FDA0003285578760000031
X axis the initial value of the optimized variable is obtained;
Figure FDA0003285578760000032
the average value of the corresponding adaptive values of P fruit flies is obtained; n represents the nth fruit fly; func (X) i ) Is X i An adaptive value function; the R band represents the rate of change of population average odor concentration; α represents an update weight; r andomValue Representing the search direction and distance.
6. The shield construction pose adjustment method for identifying hard rock thickness by using an intelligent algorithm according to claim 5, wherein in the step S63, the mapping relation in the multi-decision tree model is predicted through the data sample set II after the normalization processing, and the adaptive value function is obtained as follows:
Figure FDA0003285578760000033
in the formula: wherein yb' h Predicting the test sample values, yb, for a multi-decision tree h Actual test sample data values; f MSE Root mean square error for a plurality of test sample points; m is the number of the sample data to be tested, and h is the h-th sample.
7. The shield construction pose adjustment method for identifying the thickness of the hard rock by using the intelligent algorithm as claimed in claim 6, wherein in the step S64, the method for iterating the drosophila ejected into the drosophila elite scale is to introduce quantum operation, and the iteration process of calculating according to the quantum operation of the drosophila position and the following formula is as follows:
Figure FDA0003285578760000034
Figure FDA0003285578760000041
in the formula: n represents the nth fruit fly:
Figure FDA0003285578760000042
and u are both random numbers between (0, 1); λ is an innovation parameter;
Figure FDA0003285578760000043
is the local optimum value of the nth generation of the ith fruit fly;
Figure FDA0003285578760000044
global optimal solution for nth generation of all fruit flies;
Figure FDA0003285578760000045
represents the average of the local optimal solutions of the population of the nth generation of Elaphanita eligua.
8. The shield construction pose adjustment method for identifying the hard rock thickness by using the intelligent algorithm as claimed in claim 7, wherein the method for enabling all drosophila individuals in S65 to fly to the optimal drosophila individuals is as follows:
Figure FDA0003285578760000046
Figure FDA0003285578760000047
X axis ＝X(BestIndex)；
in the formula, ibestIndex is the serial number of the current optimal solution of the ith fruit fly; BestIndex is the individual number of the currently best Drosophila within the population.
9. The shield construction pose adjustment method for identifying the thickness of the hard rock by using the intelligent algorithm according to any one of claims 1 to 8, wherein the training process for forming the multi-decision tree prediction model in the S7 is as follows:
s71: randomly and repeatedly extracting u samples from an original learning set, and carrying out N times of sampling to form N learning sample subsets;
s72: for the N learning sample subsets, developing N decision trees;
s73: for each decision tree, assuming that S characteristic attributes are total, selecting an optimal attribute as a root node of the decision tree, and then splitting in sequence;
s74: splitting each decision tree according to the optimal attribute to form a multi-decision tree;
s75: and averaging according to the regression result of each decision tree to obtain the mapping relation between the input and the output of the multi-decision tree model.
10. The method for adjusting the pose of the shield construction by using the intelligent algorithm to identify the thickness of the hard rock according to claim 9, wherein in S8, the shield pose adjustment process according to the intrusion thickness of the hard rock is as follows:
s81: inputting the monitoring displacement and the muck density into a trained model of the multi-decision tree;
s82: the trained model of the multi-decision tree outputs hard rock invasion thickness, soft rock elastic modulus and hard rock elastic modulus;
s83: the hard rock invasion thickness is corrected as follows:
θ＝E 1 /E 2 (6)
H′＝H×δ (7)
in the formula, E 1 Is the modulus of elasticity of hard rock, E 2 Is the elastic modulus of the soft rock, theta is the elastic modulus ratio of the hard rock and the soft rock, and delta is the correction coefficient of the invasion thickness of the hard rock; h' is the corrected hard rock invasion thickness, and H is the hard rock invasion thickness predicted by the multi-decision tree.
CN202111146030.XA 2021-09-28 2021-09-28 Shield construction pose adjusting method for identifying hard rock thickness by using intelligent algorithm Active CN113836812B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111146030.XA CN113836812B (en) 2021-09-28 2021-09-28 Shield construction pose adjusting method for identifying hard rock thickness by using intelligent algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111146030.XA CN113836812B (en) 2021-09-28 2021-09-28 Shield construction pose adjusting method for identifying hard rock thickness by using intelligent algorithm

Publications (2)

Publication Number Publication Date
CN113836812A CN113836812A (en) 2021-12-24
CN113836812B true CN113836812B (en) 2022-09-20

Family

ID=78967161

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111146030.XA Active CN113836812B (en) 2021-09-28 2021-09-28 Shield construction pose adjusting method for identifying hard rock thickness by using intelligent algorithm

Country Status (1)

Country Link
CN (1) CN113836812B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114462167B (en) * 2022-03-02 2022-08-26 中车大连机车研究所有限公司 Method for evaluating axial clearance fit of double-row tapered roller bearing for railway

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106372748A (en) * 2016-08-29 2017-02-01 上海交通大学 Hard-rock tunnel boring machine boring efficiency prediction method
CN112307542A (en) * 2020-10-29 2021-02-02 南昌轨道交通集团有限公司地铁项目管理分公司 Shield construction parameter adjusting method and system for geological environment deformation monitoring

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106372748A (en) * 2016-08-29 2017-02-01 上海交通大学 Hard-rock tunnel boring machine boring efficiency prediction method
CN112307542A (en) * 2020-10-29 2021-02-02 南昌轨道交通集团有限公司地铁项目管理分公司 Shield construction parameter adjusting method and system for geological environment deformation monitoring

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
围岩流变模型参数识别的启发式-支持向量机方法;徐国文等;《现代隧道技术》;20160815(第04期);全文 *
基于IDE-OSVR-ABAQUS的岩土力学参数反演方法;卢远富等;《长江科学院院报》;20170615(第06期);全文 *
基于改进GA-SVR算法的隧道工程三维弹塑性模型参数的智能辨识;刘开云等;《岩石力学与工程学报》;20070615(第06期);全文 *

Also Published As

Publication number Publication date
CN113836812A (en) 2021-12-24

Similar Documents

Publication Publication Date Title
Jan et al. Neural network forecast model in deep excavation
KR101967978B1 (en) Apparatus for predicting net penetration rate of shield tunnel boring machine and method thereof
CN111967079A (en) Foundation pit deformation prediction method based on improved artificial bee colony algorithm and BP neural network
CN108287808A (en) A kind of general dynamic tracing sequential sampling method of structure-oriented fail-safe analysis
CN112100927A (en) Prediction method for slope deformation and soft soil foundation settlement based on GA-BP neural network
CN108204944A (en) The Buried Pipeline rate prediction method of LSSVM based on APSO optimizations
CN113836812B (en) Shield construction pose adjusting method for identifying hard rock thickness by using intelligent algorithm
CN103984788A (en) Automatic intelligent design and optimization system for anchor bolt support of coal tunnel
CN112712025A (en) Complex lithology identification method based on long-term and short-term memory neural network
CN112307542B (en) Shield construction parameter adjusting method and system for geological environment deformation monitoring
CN112035941A (en) Prediction method for surface subsidence of deep foundation pit excavation based on BAS-BP model
CN112562078A (en) Three-dimensional geological analysis prediction model construction method
CN115293316A (en) Prediction method for deep-buried thick coal seam water flowing fractured zone based on SSA-ELMAN
CN115481565A (en) Earth pressure balance shield tunneling parameter prediction method based on LSTM and ant colony algorithm
CN116050603A (en) Method and equipment for predicting and optimizing deformation of undercut tunnel based on hybrid intelligent method
CN113323676B (en) Method for determining cutter torque of shield machine by using principal component analysis-long and short memory model
CN105956715B (en) A kind of Forecast of Soil Moisture Content method and device
CN116628886A (en) Real-time optimization method for shield tunneling machine crossing construction parameters based on limited engineering data
CN112396312A (en) Open caisson sinking construction automatic decision-making method and system
CN116641688A (en) Method, system, equipment and storage medium for improving gas reservoir recovery ratio by CO2 and sealing storage thereof
CN115759203A (en) Offshore pile foundation erosion prediction method under multi-factor coupling condition
CN109117517A (en) A kind of base pit dewatering optimum design method based on genetic algorithm and groundwater model
CN113128108B (en) Method for determining diameter of jet grouting pile based on differential evolution artificial intelligence
CN108664711A (en) Anchor axial force trend method
Valetov et al. Hybrid Cuckoo Search and Harmony Search Algorithm and Its Modifications for the Calibration of Groundwater Flow Models.

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant